# Probing the Lack of Stable Internal Beliefs in LLMs

arXiv: https://arxiv.org/abs/2603.25187 · HTML: https://mlsanity.com/paper/2603.25187/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- Builds on 13 of its 23 references that are in the map.
- First posted on arXiv 26 Mar 2026.

## Builds on (most cited first, 13 of 13)

- [Evaluating Large Language Models in Theory of Mind Tasks](https://mlsanity.com/paper/2302.02083/) (2023)
- [Beyond Accuracy: Evaluating Self-Consistency of Code Large Language Models with IdentityChain](https://mlsanity.com/paper/2310.14053/) (2023)
- [Examining Identity Drift in Conversations of LLM Agents](https://mlsanity.com/paper/2412.00804/) (2024)
- [Probing the Multi-turn Planning Capabilities of LLMs via 20 Question Games](https://mlsanity.com/paper/2310.01468/) (2023)
- [Learning to Ask Informative Questions: Enhancing LLMs with Preference Optimization and Expected Information Gain](https://mlsanity.com/paper/2406.17453/) (2024)
- [Enhancing Semantic Consistency of Large Language Models through Model Editing: An Interpretability-Oriented Approach](https://mlsanity.com/paper/2501.11041/) (2025)
- [Improving Consistency in Large Language Models through Chain of Guidance](https://mlsanity.com/paper/2502.15924/) (2025)
- [Multi-Turn Puzzles: Evaluating Interactive Reasoning and Strategic Dialogue in LLMs](https://mlsanity.com/paper/2508.10142/) (2025)
- [Playing 20 Question Game with Policy-Based Reinforcement Learning](https://mlsanity.com/paper/1808.07645/) (2018)
- [Weak-eval-Strong: Evaluating and Eliciting Lateral Thinking of LLMs with Situation Puzzles](https://mlsanity.com/paper/2410.06733/) (2024)
- [Do Large Language Models have Problem-Solving Capability under Incomplete Information Scenarios?](https://mlsanity.com/paper/2409.14762/) (2024)
- [Existing LLMs Are Not Self-Consistent For Simple Tasks](https://mlsanity.com/paper/2506.18781/) (2025)
- [All It Takes is 20 Questions!: A Knowledge Graph Based Approach](https://mlsanity.com/paper/1911.05161/) (2019)

## Built on by (most cited first, 2 of 2)

- [IACM-RL: Intent-Aware Context Management and Reinforcement Learning for Complex Tool Invocation under Dynamic Intent Fluctuations](https://mlsanity.com/paper/2608.02110/) (2026)
- [From Deceptive Outputs to Deceptive Mechanisms: A Causal Framework for Language-Model Deception Research](https://mlsanity.com/paper/2609.04166/) (2026)

## Abstract (arXiv)

Persona-driven large language models (LLMs) require consistent behavioral tendencies across interactions to simulate human-like personality traits, such as persistence or reliability. However, current LLMs often lack stable internal representations that anchor their responses over extended dialogues. This work explores whether LLMs can maintain "implicit consistency", defined as persistent adherence to an unstated goal in multi-turn interactions. We designed a 20-question-style riddle game paradigm where an LLM is tasked with secretly selecting a target and responding to users' guesses with "yes/no" answers. Through evaluations, we find that LLMs struggle to preserve latent consistency: their implicit "goals" shift across turns unless explicitly provided their selected target in context. These findings highlight critical limitations in the building of persona-driven LLMs and underscore the need for mechanisms that anchor implicit goals over time, which is a key to realistic personality modeling in interactive applications such as dialogue systems.